Course details


Computer vision based asparagus classification

WS 2019 Ulf Krumnack, Axel Schaffland
M.Sc modules:
CC-MP-SP - Study Project
CS-MP-SP - Study Project
KOGW-MPM-SP - Study project
Fri: 8-10

Prerequisites: Basic knowledge of neural networks and some experience with a deep learning framework (e.g. TensorFlow) are requisites for this course. Having attended the class „Implementing ANN with TensorFlow“ is a good basis for joining the study project. Abstract: An essential step in commercial asparagus cultivation is the sorting of the harvested stalks into different commercial classes. Depending on size, shape and colour, the individual spears are assigned to a quality class which determines the price of the spears. Nowadays, this task is usally done automatically, with modern asparagus sorting machines achieving a throughput of up to eight spears per second. However, the accuracy of such machines is often unreliable, making manual resorting is necessary, causing non-negligible costs. The study project will investigate how techniques from computer vision, both classical and deep learning based, can be applied to this task to improve classification results. The project will be supported by a local asparagus farm, providing training data and allowing to evaluate the proposed solutions in a real environment.